GLOSSARY
Model Risk Management
The discipline — codified by OCC SR 11-7, EBA, and BIS guidance — of identifying, measuring, monitoring, and controlling the risk of using statistical or AI models to make decisions. Standard in finance; applicable to trade-AI.
Model Risk Management (MRM) is the framework banks use to govern any model whose output drives a decision: credit scoring, sanctions screening, capital adequacy, fraud detection. The canonical reference — OCC SR 11-7 (2011) — covers model definition, development standards, validation, ongoing monitoring, and governance, with parallel EBA, ECB, and BIS guidance in Europe.
Why it matters
Trade-AI agents that decide who gets credit, which shipment is screened, or which sanctions hit clears are models in the SR 11-7 sense — same expectations apply. Adopting MRM-style controls early (model inventory, validation, drift monitoring) is what makes a trade-AI deployment defensible to financial-services counterparties and supervisors.
Related terms
- High-risk AI System
- Explainability
- Validation
- Model Inventory